A REVIEW OF THE A . T. KEARNEY APPLICATION OF THE WARTON EFA AUTOMOBILE DEMAND MODEL

TO ASSESS THE IMPACT OF ENVIRONMENTAL REGULATIONS

Barbara C . Richardson

David C . Roberts

Kent B . Josce lyn

October 1979

Po l icy Analysis Div i s ion Highway Safety Research I n s t i t u t e

The U n i v e r s i t y o f Michigan Ann Arbor, Michigan 48109

ACKNOWLEDGMENT

Development of this independent paper was supported b y an

unrestricted grant from the Motor Vehicle Manufacturers Association.

The assistance and advice of colleagues, particularly Prakash 8. Sanqhvi

and Daniel B. Suits, is gratefully acknowledged.

ABSTRACT

This paper r e p o r t s a rev iew on t h e use of t h e Wharton EFA

Automobile Demand Model by A.T. Kearney, Inc. in i ts study of the

economic impacts of environmental regulations on the automobile industry,

This review is based on the Kearney draft final report to its sponsor, the

Environmental Protection Agency (EPA); the Wharton EFA Automobile

Demand Model documentation; and a previous analysis of the Wharton

EFA auto model by the Highway Safety Research Inst i tute (HSRI) of The

University of Michigan.

This review examines Kearney's modifications and extensions of the

Wharton EFA auto model, and how the use of tha t model af fected the

Kearney analys is . In addi t ion, the model's simulation of the most

probable regulatory scenario is briefly interpreted.

The findings of the review are:

Kearney modified the Wharton EFA auto model to forecast the effects of EPA regulations on new car sales; size-class market shares; gasoline consumption; the profitability of AMC, Chrysler, Ford, and GM; and automobile industry employment.

e The Kearney ana lys i s in i t s d r a f t form contains an informative description of the motor vehicle transportation industry but it fails to adequately: (1) explain the Kearney modifications of the Wharton EFA auto model, ( 2 ) detai l the limitations of the Wharton EFA auto model, and (3) interpret the forecasts of the Wharton-Kearney model simulations.

e The Kearney modi f ica t ions to the Wharton EFA auto demand model did not c o r r e c t any of t h e s e r i o u s weaknesses of the original Wharton EFA auto model with regard to policy applications. Furthermore, the Kearney add i t ions of t h e profi tabil i ty and automobile industry employment equations have weaknesses tha t det ract from t h e limited benefits achieved by the inclusion of those equations. Thus, the W harton-K earney model cannot be r e l i ed upon i n policy analysis applications designed to estimate the economic impacts of environmental regulations.

1.0 INTRODUCTION

This paper is a brief review of how A. T. Kearney, Inc. used the

Wharton Econometric Forecasting Associates Automobile Demand Model in

studying the economic impacts of environmental regulations on the

automotive industry (Kearney 1978). This review is based solely on an

examination of the Kearney draf t f inal repor t , since the program computer code and the program documentation were unavailable. This

review, performed by the Policy Analysis Division of the Highway Safety

Research Institute (HSRI) of The University of Michigan, was part of the

study entitled, "Analytical Study of Mathematical Models: General Policy

Studies." The reader will more easily understand this review if he is

familiar with the Wharton EFA automobile model. Further information on

that model is available in the original model documentation (Schink and

Loxley 1977) and i n an analysis of tha t model b y Golomb, Luckey,

Saalberg, Richardson, and Joscelyn (1979).

The Kearney study was sponsored by the U.S. Environmental Protection

Agency (EPA). Its objective was to examine the economic impacts of

various pollution standards on the automotive industry.

There are three parts to the Kearney Report. The first part describes

the automotive manufacturers (American Motors Corporation !AMC 1,

Chrysler Corporation, Ford Motor Company, and General Yotors

Corporation !GM1, parts suppliers, and supporting service industries) and

the characteristics of automobile demand, including abstracts of various

automobile demand models. Part two presents the Kearney estimates of

the incremental costs to both the automakers and their customers due to

potential EPA emissions regulations in future years. Part three presents

the Kearney estimates of economic impacts of the emissions abatement

requirements,

The portion of the Kearney analysis relating to the impact of EPA

passenger car regulation is based on simulations of the R earney-modified

1977 version of the SVharton EFA Automobile Demand Model. This review

addresses the effect of using the Wharton EFA automobile model i n the

Kearney analysis.

The Wharton EFA Automobile Demand Model was selected by Kearney

as its major analytic tool, because its capabilities most closely matched

the s tudy needs. The Wharton EFA Automobile Demand Model was

developed in 1976 under the sponsorship of the Transportation Systems

Center (TSC) of the U.S. Department of Transportation. The project was

initiated specifically to provide that federal agency and others with a

better analytical tool for investigating the potential impacts of proposed

policies and regulations on the motor vehicle industry and on the economy

in general.

This paper is organized as follows: Section 2.0 describes the Kearney

modifications to the Wharton EFA automobile nod el. Sect ion 3.0

examines the impact of the use of the Wharton EFA Automobile Demand

Model on the results of the Kearney analysis. Section 4.0 discusses the

model's simulation results of the most probable regulatory scenario.

Section 5.0 presents the findings of this review.

2.0 THE KEARNEY MODIFICATIONS TO THE WHARTON EFA AUTOMOBILE DEMAND MODEL

Kearney modified the Wharton EFA auto model so that i t could

s imula te the e f f e c t s of s t a t i o n a r y - s o u r c e and mobi l e - source

pollution-abatement regulations. The objective was to estimate the

impact of these regulations on gasoline consumption and auto industry

employment and profitability. The two mobile-source regulatory scenarios

simulated by Kearney were based on U.S. House of Representatives Bill

61 61. They include alternative emissions standards for hydrocarbons,

carbon monoxide, and oxides of nitrogen. These scenarios are presented

in Table 2-1. Each of ,these regulatory scenarios is simulated under two

assumptions concerning auto industry behavior: an optimal fuel-economy

assumption and an optimal cost assumption. Kearney reports that the

most probable scenario is Case A with an opt imal fuel-economy

assumption.

The Kearney analysis examines the impact of regulations on new car

registrations, revenues of foreign and domestic automakers, average cost

per mile, total gasoline consumption, U.S. auto industry employment,

market shares of five size classes, total domestic market share, total

regulatory cost per new car (including mobile-source costs as well as the

effect of regulations on fuel economy), average fuel economy (both new

and i n total), and the gross operating margins for the Big Four domestic

automakers. Since the original Wharton EFA auto model is not capable

of forecasting these effects, Kearney modified the model.

The Kearney modifications to the Wharton EFA auto model do not

slter its basic structure. The modifications are divided into two groups:

(1) "Front endt1 modifications designed to input to the model the mobile

and stationary source costs, maintenance cost changes, and fuel efficiency

changes owing to the EPA regulations; and ( 2 ) "Back end1' modifications tha t use the model's outputs to develop es t imates of revenues,

TABLE 2-1

MODEL YEAR

A I R POLLUTION CONTROL SCENARIOS FOR LIGHT-DUTY MOTOR VEHICLES

HC/CO/NOx (grams/mile)

CASE A CASE B ROGERS' BILL (H.R. 6161) ROGERSt BILL (H.R. 6161)

1.5/15/2.0 1.5/15/2.0

Notes: 1. Case A assumes t ha t the EPA Administrator w i l l decide i n 1980 t h a t more s t r ingen t NOx standards w i l l not be required f o r ca rs produced i n the 1983 model year and t he r ea f t e r .

2 . Case B assumes t h a t the EPA Administrator w i l l decide i n 1980 that more s t r ingen t NOx standards w i l l be required f o r ca rs produced i n the 1983 model year and t he r ea f t e r .

3. NOx standard would be subject t o change a t the d i sc re t ion of the EPA Administrator, consis tent with clean a i r qua l i ty .

Source: Kearney 1978, Part 2 , pp. 11 -2 .

profitability, employment, and gasoline consumption. These modifications

are illustrated in Figure 2-1.

Front-End Modifications

1. Fuel-Ef ficiency Change Factors

Fuel economies by size class of vehicle are calculated in the Wharton EFA auto model. Kearney develops projections of fuel

economy changes resul t ing from emissions s tandards and

incorporates them into the model through use of a multiplicative

factor.

2. Maintenance-Cost Change Factors

The increases in maintenance costs resulting from the E P A

standards enter the model as additional repair costs. These

costs, presented in current dollars, are the total costs over the

l i f e of the new automobile. These costs are appropriately

discounted and deflated to constant-dollar terms to be consistent

with the other costs in the Wharton EFA auto model.

3 . Emission Abatement Costs

The incremental costs resulting from the E P A regulations on

mobile and stationary (both manufacturing and supplier facilities)

sources of pollution are input to the Wharton EFA auto model by

attaching three variables (one for each source) to each of the

eight stripped (without options) base prices by size class. These

costs are input i n current dollars to be consistent with the

Wharton E F A auto model. The mobile source costs are assumed

to be passed along to the consumer without a markup. However,

the cost of stationary-source pollution control is assumed to be

passed along to the consumer with a fifty percent markup (Note:

the Kearney documentation contained contradictory statements

concerning this markup; an appendix stated no markup was used,

while the text stated in several places that the markup was fifty

percent ).

4 . Diesel Penetration Factor

One of the sensitivity tests considered in the Kearney report is

the impact of the penetration of the use of diesel engines into

the new automobile market. This t e s t was regarded as an

al ternat ive to the regulatory scenarios (i.e., Cases A and B)

simulated. This test simulates increased diesel penetration a t the

rate of five percentage points per year, beginning in 1981, up to

a total of twenty-five percent of new car sales in 1985. The

twenty-five percent figure is then held constant through 1990.

To perform this test, a variable containing the fraction of

diesel penetrat ion is a t t ached to each of the eight base

price-by-size-class equations to adjust the base prices for the

higher initial costs of diesel-powered automobiles, In addition,

diesel penetration is also made to affect maintenance costs and

f u e l e c o n o m i e s . G a s o l i n e c o n s u m p t i o n , under t h e

highdiesel-penetration scenario, appears to include both gasoline

and diesel fuels. Neither the price nor the consumption of diesel

fuel are considered in the sensitivity test.

2.2 Back-End Modifications . .

1. Gasoline Consumption

Kearney estimated gasoline consumption by dividing the forecast

of total vehicle miles traveled by the average total fleet fuel

economy. The gasoline consumption equation is, however, a

relatively minor addition to the Wharton EFA auto model, since

the independent variables are already part of the output.

2 . Profitability Equations

Kearney developed a profi tabi l i ty equation for each major

domestic manufacturer: AMC, Chrysler, Ford, and GM. The

dependent variables for the equation, as reported in the Kearney

report, is gross operating margin. That is defined as operating

earnings (before deduction of depreciation, depletion, interest, and income taxes) as a percentage of net sales (Kearney 1978, Part

111, pg. v). However, the HSRI staff, using Kearney-supplied income statements for the automakers, determined that the data

used for regression purposes has operat ing income a f t e r

depreciation and amortization as a percentage of new sales. Net

sales includes the sales of all products by the companies, not just

automobile sa les in the United States. Three of the five

independent variables are the proportion of the manufacturersf

sales in (1) the subcompact and compact submarket, ( 2 ) the

intermediate submarket , and ( 3 ) t h e ful l -s ize and luxury

submarket . The o the r independent variables are capacity

utilization and the manufacturer's market share . Capaci ty

utilization is the manufacturer's new registrations relative to a

five-year moving average. This is a percentage. The market

share var iable is an index of the sales-weighted average of

market shares in each size class. That is, the market share

index values are calculated by multiplying a firm's market share

in a size class times the fraction of the firm's dollar automobile

sales comprised of that size class, and then summing over all

size classes. The profitability equations were estimated using a

generalized least-squares technique applied to annual data from

1947 to 1976. These results are shown in Table 2-2.

The estimated coefficients of sales shares support the general

belief that Ford, GW, and Chrysler profit more from the sales of

larger cars. The A M C estimated coefficients are an exception.

However, AMC has become primarily a small car producer, and

the estimated coefficients indicate that AMC profits more from

small car sales than from the sales of larger cars.

Capacity utilization is expected to be positively related to

prof i tab i l i ty . The values of t h e c o e f f i c i e n t s and t h e

corresponding t - s t a t i s t i c s show that capacity utilization is

significantly important in estimating the profitability of t h e

automakers . The market share index is shown to be not

significantly different from zero at the ten percent level.

The standard errors of regression are high when compared

with the means of the automaker's operating margins over the historical period. Recalling that the dependent variables are

percentages, these means are 15.46, 9.346, 4.715, and 1.35 for G M ,

0 a - .i Ln h Y l - D l n crib z 7 2 k w g Y12 U 0 z c Y m a .. m a u ) Y

k k V m u w 0 I.. h = ; ? g 3 Y m m U m k .d 3 4 k a

2 g c a C *I .- e, 3 1 2 3

Y1 .i u) a w m - 0 k j h m a h 3 * 0 a 3 m - m

.i Y * u : Y 5 2 2 .:

L L *J s m M U

: g m e 3 2 ; $ b) .4 c a 3 r a m 1 MCJZ .2 x m u acu) m

- Q ) 0 J E 3 a 0 0 1 r * m a m i k m :: u s 0 i u).r( m p

c I n - m

m u * c . - Y I 0 - a .- m o - In .d V l Y 3 Y m m m L( k u ? m YE = m U <

Ford, Chrysler, and AMC, respectively. This indicates that the

forecast levels of operating margins for the automakers are likely

to be qui te f a r off the mark, particularly in the cases of

Chrysler and AMC.

The profitability equations as specified by Kearney imply that

the costs of the EPA regulations have only an indirect effect on

the profitability of the automakers, that is, the costs are filtered

through the independent variables of the equations, This mav be

t r u e i n some f i rms , but i n other firms the regulations may

impose direct and additional hardships. For example, a firm that

lacks capital may have difficulty raising the additional funds for

stationary pollution-control equipment, or a firm may have to

sh i f t key personnel into management of the pollution-control

program. These types of profitability losses are assumed minor

by Kearney. 2

The profitability equations have low R statist ics for time

series. This may be caused by Kearneyfs e s t ima t ing t o t a l

profitability of the firm based on what occurs in the passenger

car market. The major automakers, however, have significant

interests in such areas as trucks, buses, tractors, qlass, financing,

locomotives, and replacement parts, plus overseas operations,

Considering the breadth of those interests, it is not surprising 2 that the R s and some of the t - s t a t i s t i c s a r e low. These

equations would be useful for estimating the effects of the

changes in profitability resulting from changes in the auto market

i f the auto market profit source were independent of the other

sources of profit. However, this is probably not the case.

Events affecting profits from the sales of new cars will likelv

affect the automaker's other sources of profit. Therefore, the

equation excludes re levant var iables and produces biased estimates of the coefficients.

To l i n k the outputs of the Wharton EFA model to the sales

shares used by the profitability equation, Kearney developed a

"solution algorithm." The solution algorithm produces product

mixes (the fraction of a manufacturer's output i n each class)

given the firm's market shares (manufacturerls share of total

industry output) and the size-class market shares. The size-class

market shares are outputs of the Wharton-Kearney model. The

firm's market shares are exogenous to the Wharton-Kearney

model. The firmsf market shares input to the Wharton-Kearney

model are in Table 2-3.

3 . Automobile Manufacturing Employment Equation

Based on a Cobb-Douglas production function and the

hypothesis that employment demand in the short run will differ

from its long-run equilibrium value, Kearney developed an

automobile manufacturing employment equation. Kearney

estimated that relationship to be in the following form:

R' = , 507 S . E . = .0880 D.W. = 1.85

where

Lt = employment in period t

o t = output in period t

Wt = real wage rate in period t The output and real wage variables have the long-run effect on

employment. The lagged employment variable moderates the

short-run impacts of changes in the other independent variables.

The HSRI staff have the following observations and comments

about the employment equation. First, the dependent variable as

used for regression purposes is production worker employment in

Standard Industrial Classification (SIC) 371. This class contains a

substantial number of nonpassenger car production workers. SIC

371 includes workers producing trucks, buses, tractors, motor

vehicle parts and accessories, and truck trailers. In 1974, parts

and accessories production workers accounted for almost half of

TABLE 2 - 3

ASSUMPTION OF MARKET SHARES OF BIG FOUR FIRMS

1975-1990

GM - FORD - CHRY SLER AMC -

Source: Kearney 1978, P a r t 3, Tab le D - 1 .

all SIC 371 production workers (U.S. Department of Labor 1976).

Thus, assuming SIC 371 employment depends entirely on sales of

new cars is a very rough approximation. Second, it should be

observed t h a t the only way t h a t t e c h n o l o g i c a l c h a n g e

(productivity) is accounted for in the equation is through the real

wage variable. Third, the equation would be better specified i f

the dependent variable were worker-hours or the equivalent

number of full-time workers, since the nature of the industry

dictates overtime and cutbacks on an irregular basis.

The forecasts of the real wage rate (i.e., the future course of

productivity) are based on a time-trend equation estimated over

the period 1948 to 1976 and are assumed by Kearney to be

determined exogenously to the Wharton E F A auto model.

Therefore, the impacts of the EPA regulations are transmitted

solely through changes in the output variable, which is new car

registrations as forecast by the Wharton EFA auto model.

All the coefficients in the estimated equation are significantly 2

different from zero, but the R of ,507 is relatively low for a

time-series equation. The standard error of regression as a

percentage of the mean of the dependent variable is 22.46%.

3.0 LIMITATIONS IMPOSED ON THE ANALYSIS BECAUSE OF USE OF THE WHARTON EFA AUTOMOBILE DEMAND MODEL

An analysis based on an econometric model is subject to the model's

limitations. Thus, the conclusions from the Kearney study of the impact

of passenger car regulations are dependent on the characteristics of the

Wharton EFA auto model.

3.1 The Wharton EFA Auto Model as a Policv Tool

Golomb et al. (1979) found the Wharton E F A au to model to be

generally insensitive to changes in new car prices, automobile use costs,

and production input costs. This insensitivity may be reasonable in the

case of changes in some costs such as insurance, parking and tolls, and

possibly repair costs, but inappropriate i n other cases such as purchase

price and fuel costs. They note that this insensitivity is particularly true

in the long run.

The source of these insensitivities lies i n Wharton's methodology.

Golomb et al. (1979) conclude:

A fundamental weakness of the model involves the price elasticities in the desired stock and share equations. Desired stock and desired stock shares by vehicle class are critical to the operation of the model either in long-term forecasting or policy analysis. Yet these critical equations are estimated over a 1972 cross-section of s ta te data. The cost of owning and operating a car will vary minimally across states, and the variation that does exist will not be due to fundamental differences i n the price of a car, the level of federal taxes, and so on. In these circumstances, one can have very l i t t le confidence i n the estimated cost or price elasticities. They simply do not reflect responses to price variations that would be observed over time as the result of changes in policy. In effect, this rules out the possibility of using the model with any confidence in a number of interesting policy simulstion experiments. (p. 205.)

I n addition, the Wharton EFA auto model's new-car-sales-demand

equation spec i f ica t ion has demand as a funct ion of the ratio of

current-year prices to the prior-year prices. This specification produces

large first-year impacts due to price changes but no long-run impact.

The long-run impact of the price changes occurs as a result of changes in

the desired stock due to changes in the capitalized cost per mile. The

round-about route produces small long-run impacts of price changes.

3.2 The Effect of Using Wharton EFA Automobile Model Outputs

Given the problems mentioned in the previous section, the reliability

of the specific Wharton output used by Kearney is questionable. Table

3-1 lists the Wharton model outputs used by Kearnev, along with the

respective error statistics that were produced by HSRI staff when they

exercised the model over the period 1960-1974. The interpretation of the

error statist ics is straightforward. The forecasts of new car sales, VMT,

and average MPG are reasonably accurate. The relativelv high errors for

market shares indicate that the Wharton EFA auto model inadequately

represents the determinants of new car registrations by size-class market

shares.

These statist ics and other conclusions reported by Golomb et al. have

implications for the reliability of the Kearney simulation results. These

implicat ions as well as the HSRI staff's analysis of the particular

requirements of the Kearney study are discussed below.

1. Impact of Fuel Efficiency Changes on Wharton-Kearney Model

Outputs

The impacts of EPA-caused fuel economy changes on the

model's output variables are transmitted through the capitalized

cost per mile variable. Capitalized cost per mile is a critical

variable in the Wharton EFA auto model. It includes finance

charges, gasoline, insurance, t ire, parking and tolls, and motor

oil costs as well as the purchase price of new cars (including

t axes ) , Capi tal ized cost per mile is a determinant of the

desired automobile stock and desired market shares by size class

of new car sales. These latter two in turn are determinants of

new car sales, both in total and by size-class market share.

The point of this discussion is that changes in fuel efficiency,

both in the short and long runs, will have minimal impact on the model's output variables due to the structure of the model.

That is, the importance of changes in fuel economy in the

purchase decision is not high.

Profitability Forecasts

As noted in the previous section, the profitability forecasts

are dependent on the size-class market shares as estimated by

the Wharton EFA auto model. The error statistics presented in

Table 3-1 indicate that the Wharton EFA auto model produces

relat ively inaccurate and unreliable estimates of size-class

market shares. Since the Wharton EFA auto model inadequately

represents the determinants of market shares, any use of these

forecasts for policy analysis purposes is likely to produce

incorrect conclusions.

Profitability changes due to shifts in size class com~osition

of sales caused by EPA regulations cannot be regarded as

reliable if the model does not (a) represent what determines market shares and (b) have reasonable estimated cost or price

elasticities.

Gasoline Consumption

Kearney forecasts gasoline consumption by dividing total V M T

by average fleet fuel economy, both as generated b y the

Wharton EFA auto model.

Average fuel economy is critically dependent on the stock of

cars by class and vintage. Since the number of cars added to

the stock in any given year is relatively small, average fuel

economy as forecast by the model in the short run is dominated

by the stock of cars already in existence. However, in the long

run, forecasts of average fuel economy will be dependent on the

s tock of cars as forecast by the Wharton EFA auto model.

Since the model forecasts new car size-class market shares with poor reliability, long-run forecasts of average fleet fuel economy

are equally unreliable.

TABLE 3-1 .

ERROR STATISTICS WHARTON EFA AUTOMOBILE DEMAND MODEL

SIMULATION OVER THE WITHIN-SAMPLE PERIOD 1960-1974

VARIABLE

VMT/ Family Unit

Average F l e e t Fuel Economy

New Car Reg i s t r a t i ons

Market Shares 3

Subcompact

Compact

Midsize

F u l l

Luxury

Subcompact and Compact

Midsize

Fu l l and Luxury

Notes: 1. RMSE i s t h e average e r r o r of t h e p red ic t ed va lues and i n d i c a t e s t h e accuracy of t h e f o r e c a s t .

2 . % RMSE is the RMSE a s a percentage of t h e mean of t h e a c t u a l va lues of t h e v a r i a b l e s over t h e f o r e c a s t per iod , i . e . , 100 x RMSEIMean.

3. While t h e hharton EFA Auto Model f o r e c a s t s f i v e s i z e - c l a s s market shares , t h e Wharton-Kearney vers ion com- b ines t h e subcompact and compact sha re s , and t h e f u l l and luxury shares f o r use i n t h e p r o f i t a b i l i t y equat ions. Combining t h e shares reduces t h e e r r o r l e v e l s cons ider - ab ly .

Forecasts of VMT are dependent on forecasts of size-class

market shares, for the VMT equation reflects variations i n

mi l eage due to changes i n the age composition of the

automobile stock. The VMT estimate is also dependent on the

average fuel economy as forecast by the model. In the short

run, VMT is relatively insensitive to the policies simulated by

Kearney, since those policies are transmitted through the model

by changes in size-class market shares. But, as with fuel

economy, the prediction of VMT in the long run is critically

dependent on the poor estimates of market shares generated by

the Wharton EFA auto model.

To summarize, both VMT and average total fuel economy as

estimated by the Wharton EFA auto model are insensitive i n the

short run to the policies simulated by Kearney. In the long run,

the policies1 effectiveness is t ransmi t ted to the gasoline

consumption estimates through the market shares and new car

sales forecasts. Since the size-class market shares forecasts are

unreliable, the policies1 impact on gasoline consumption cannot

be accurately measured.

4 . Automobile Industry Employment

In forecasting the employment effects, the only Wharton EFA

auto model output used b y Kearney is the new car sales

(registrations). Over the historical period, it has been shown by

Golomb et al. that new car sales is forecast with an RMSE of

820 ,000 units or a percentage error of 9.5%. The trends of new

car sales (i.e., the upturns and downturns of sales) are predicted

fairly well. However, the other important requirement is that

the cost and price elasticities reflect reality. As indicated

previously, these elasticities are estimated over a questionable

data set, i.e., cross-section data across states. Thus, use of the Wharton-Kearney model for policy simulations that affect prices

or costs are likely to produce inaccurate results.

5 . Foreign and Domestic Revenues

The W harton-Kearney model forecasts both domestic and

foreign revenues from the sales of cars i n the U.S. As Kearney

notes in its report, the foreign/domestic 'split is exogenous to

the version of the Wharton EFA auto model used, and this

limitation must be considered when evaluating the effects of

various policies.

In the Wharton EFA auto model, foreign makes compete in

the subcompact, compact, and luxury car submarkets. The

Kearney foreign share of those size classes are 51%, 5%, and

9096, respectively. However, it should be noted that Schink and

Loxley (1977) report these foreign shares to be 52%, 7%, and

12% respectively for the period 1976-2000. Thus, it appears that

Kearney has either changed the values of the split parameters

substantially or else has some typographical errors in its report.

Kearney performed sensitivity tests on these externally set

parameters by simulating a regulatory scenario under t h r e e

groups of parameters. These groups are (1) original Wharton

EFA auto model parameters noted previously, ( 2 ) original

parameters increased by lo%, and ( 3 ) original parameters

decreased by 10%. Changing these parameters b y 10% does not

imply that the foreign manufacturerst share of the U.S. market

changes by 10% since the foreign manufacturers are assumed by

Wharton EFA to have no ent r ies in the intermediate and

full-size classes. The test results showed that the model's

fo recas t s a r e changed b y insignif icant ly small amounts.

However, a conclusion that the foreign/domestic split has only

insignificant impact on the Wharton-Kearney outputs variables, including profitability, cannot be considered reliable without

additional inform ation.

In addition, it should be noted that the import car market

has been changing substantially in recent years. There is now

foreign car penetration i n the mid-size class market. This is

not considered in the Wharton EFA auto model, and adds to the

uncertainty of the Wharton-Kearney results.

4.0 DISCUSSION OF A SELECTED WHARTON-KEARNEY MODEL SIMULATION

To facilitate an understanding of the issues raised in previous sections,

this section presents a brief discussion of the results of a simulation of

regulatory scenario A under the assumption of optimal fuel economy. The

simulation discussed here is one of many reported by Kearney and is

indicative of other simulations. Table 4-1 contains the simulation results.

The upper portion of the table shows the simulation values of the output

variables. The lower half of Table 4-1 indicates (in unit or percentage

terms) the impact of the regulations on a baseline case involving no

additional abatement costs. The major results are discussed below.

1. New Car Sales

As expected, the impact on new car sales of the changes in

price are most dramatic in the year the price increases occur.

However, once the price stops rising (1982), sales return to close to

normal. The positive sales increase over the baseline in 1982 is

counter intui t ive, given tha t buyers that year are faced with

pollution abatement costs $300 higher than in the baseline case.

2. Domestic and Foreign Revenues

In evaluating these values, recall that the foreign shares of

three size classes are set exogenously. The increases in revenues

are a result of the increases in the prices of automobiles due to

the pollution control equipment.

3. Employment - Auto Industry

The employment f o r e c a s t s reflect the construction of the

employment equation (the prior-period employment variable) and the

forecasts of new car sales. In 1980, only a partial impact of the

sales decrease occurs. I n 1981, the second consecutive year of

depressed sales, employment drops even further. I n 1982, the

prior-period employment variable shows i ts effect as employment

I l l h h h h h h - - w

o n * n h n -hN ... h n e -IY

w h o N h P (Cm- 0 . .

C ' t * - r4 U 1 H n e 3 hEq .. n n m a H

h h h h e n ?PC . . (n H P - (Y

C r4 8 h N P

? ? ? r r4 E a m

e n s t w n 9 9 9 n-P - 3

d 5 2

5 1 C P : D C C t u r w w a m u 5 a w I i* Z k C - C

3 W C W L U Z C W -L 'CI (L-0 e =

i m c C W C t - r - - 2 r C S

- -n Chn Q rd 0. . . . n o n

1

cnr) O e 3 * m e . - . nor,

I

- f

Z C r E 3 I C C - < Z O E - c

C C o e r 2:: - y o C L Z co: a w m c L m P : S G +

tr c=sl c+cn = s z t r = g g - - = L U C

0 0 nn -- . . e n

O t P P

?: 7 b7

h a 0 ri ? P1 o n i

N- h d n* . . e n

h N h..

? ? n- I

-0

Z ? n o I I

u U 2 , W C:

g - CI

I

LL( C E I Y u L s Y W 0 a

I I W & > - r w C c a

U C - c- ud,

LI;

o h C O 9 9 0 0

r h G O 0 0 . . O C

0 = O = C C . . 0 C

g g 9' 0 0

C P C C 0 0 . . 0 0

1

o n 0 0

9 9 00 I

W E z u E

'tt k L

C II

z - - W 5: ,s, a=

C C = U C C

k c c C U i = = Z E La;;

O t O C O O 0 0 : . . c e o

I

c m o 0 0 0

C.99 0 7 0

C O O O C C 999 0 0 0

I

00.0 O O C 4 9 9 0 0 0

i

8 P . C 0 0 0 o c q . . O O C 1 I

o n 0 e c o c o q . . C O O

1

W. Y c = s Z z g Lfi c

L;'= UUD: k - m m * a ' ,B A-z - - = = E lb-'

-n- O l h o q * . . O b R I n h

N I

3 h N 01- C t - ... C P T I n h

N I

- N O C C C I 0 C i h ... o m n I nh u I

oe- 00:

??-. O b h

I n h rd I

C m h 0 &7 + o - n . . . c P.P I H h

M I

C H I C N P o t - ... 0-C l -h - t

- - 0 0

u u I ' hr; h h v u

LLC c e U U

3 % L ; i

W L Z c\\ C r - = o m U C C

UU

ECLL' - 6 2 ???Z t S E 2aC. C c

Y C .A .c - - C 'j - - t 5 E

drops to 1.0% below the baseline case, even though sales are .4%

higher than the baseline case. After 1982, the long-run impact of

new car sales dominates the employment forecasts and is relatively

minor.

4. Size-class Market Shares

Trading down occurs as the capitalized cost per mile increases.

The extent of the trading down is minor; a t the most the shares

change by .9 percentage points, or less than 5%. The reasonability

of such minor changes is difficult to evaluate. However, as noted

previously, the reliability of the market shares forecasts is not

high, which leads the HSRI staff to have serious reservations about

these results.

5. Total Fleet In-Use MPG

As expected, the impact of the regulatory scenario on MPG is

verv gradual and increases over the long run due to both the

absorption of new cars with higher MPG than the baseline and the

trading down that occurs relative to the baseline. The impact in

the long run is again dependent on the size-class market shares

forecasts.

6. Operating Margins of Firms

The impact of regulations on profitability margins is minor.

AMC is shown to have some slight profitability increases due to

the regulations, while Chrysler is shown to have negative

profitability margins in 1981, 1984, and 1985 because of the EPA

regulations. Conclusions drawn on these results are subject to

error. In addition to misspecification, the profitability equations as

estirnat ed b y K earney have relatively high standard errors of

regression. This indicates that the levels of profi tabi l i ty as

forecast by the Wharton-Kearney model are not precise. Therefore,

the conclusion that a firm may be in a loss situation because of

the regulations could be drawn, but it would be unwarranted. On

the other hand, a firm forecast to be earning a profit may actually

be i n a loss situation. This is not well explained in the Kearney

report.

5.0 FINDINGS

The HSRI review of a portion of the A. T. Kearney analysis that was

based on exercising the Wharton-Kearney model produced the following

major findings.

a Kearney modified the Wharton EFA auto model to forecast the

effects of EPA regulations on new car sales, size-class market

shares, gasoline consumption, the profitability of AMC, Chrysler,

Ford, and GM, and automobi le indus t rv employment . These

modifications provided innovative additional capabilities to the

Wharton EFA auto model. However, the Kearneyfs profitability and

employment equa t ions were poorly spec i f i ed as a result of

simplifying assumptions.

a While t h e Kearney analys is in i t s d r a f t form con ta ins an

informative description of the motor vehicle transportation industry,

i t fails to adequately (1) explain the Kearney modifications of the

Wharton EFA auto model, (2 ) detail the limitations of the IVharton

EFA a u t o m o d e l , a n d ( 3 ) i n t e r p r e t t h e f o r e c a s t s of t he

Wharton-K earnev model simulations.

The Kearney modifications to the Wharton EFA auto demand model

did not correct any of the serious weaknesses of t h e or ig inal

IVharton EFA a u t o model with regard to policy applications.

Furthermore, the Kearney add i t ions of t h e p ro f i t ab i l i t y and

automobile industry employment equations have weaknesses that

detract from the limited benefits achieved by the inclusion of those

equations. This policy application of the Wharton EFA auto node1

relies heavily upon that model's cost and price elasticities and its

representation of the determinants of market shares. However,

these characteristics of the Wharton EFA auto model a re i ts major deficiency in policy applications. Therefore, the Wharton-Kearney

model cannot be relied upon in policy analysis a~pl ica t ions designed

to estirnate the economic impacts of environmental regulations.

BIBLIOGRAPHY

A. T. Kearney, Inc. 1978. Economic impact analysis of E P A regulations on the automotive industry. Draft final report. Chicago, Illinois.

Golomb, D.H.; Luckey, M.M.; Saalberg, J.H.; Richardson, B.C.; and ~ o s c e l ~ n , K.B. 1979. - kn analysis of the Wharton EFA automobile demand model. Preliminarv draft. Ann Arbor, Michigan: UMI Research Press, an imprint of university Microfilms 1nternationG.

U.S. Bureau of t he Budse t , Of f ice of Stat ist ical Standards. 1967. . . - .

Standard industrial' classification manual. Washington, D . C .: U.S. Government Printing Office.

U.S. Depa r tmen t of Labor. 1976. Emplovment and earnings: U.S. 1909-1975. Washington, D.C.: Government Printing Office Bulletin 1312-10.